Here is an interesting study that suggests your average diet contains both anti-cancer and cancer causing compounds and the effect sizes cancel out. For example bacon increases your risk of cancer and onion decreases your risk. So eat bacon and onions and you’ll be perfectly healthy.
http://ajcn.nutrition.org/content/early/2012/11/27/ajcn.112.047142.abstract

I was thinking about this paper and here’s the twist. Because it measures your relative risk of cancer to the average population, it could be that most of the foods cause cancer just to varying degrees. Because the average population eats most of the foods, your relative risk decreases if you eat onions instead of bacon, but perhaps this is just because onion is carcinogenic but less so than bacon. It’s weird to think about it that way, but here’s a hypothetical example to explain it. Imagine that everyone in the world spent an hour a day either smoking, getting an xray or a mixture of both (but not both at the same time!). Some people did more smoking and some people did more xrays it was up to you! Then we did a similar study to the one above and looked at the population and compared their cancer rates to how long they spent either smoking or getting an xray. This study would find that smoking reduces your relative risk of cancer and we would come to the conclusion that everyone should smoke! Because most people eat and drink similar amounts if you ate food A over food B it could seem to reduce your risk of cancer (based on the population) but physiologically food A could be increasing your risk of cancer but just by less than food B. Also the reverse could be true. Most foods physiologically inhibit the process that induce cancer (DNA mutation) just to varying degrees.